NVIDIA H100 SXM vs NVIDIA H20 (2024): Spec Comparison & Buyer's Guide
In AI infrastructure selection, NVIDIA H100 SXM and NVIDIA H20 (2024) are two accelerators frequently compared. This article contrasts them item by item — architecture, compute, memory, power, and release cadence — to help you quickly judge which fits training or inference workloads.
Spec Comparison Table
| Vendor | NVIDIA H100 SXM | NVIDIA H20 (2024) |
|---|---|---|
| Vendor | NVIDIA | NVIDIA |
| Architecture | Hopper GH100 | Hopper |
| Process | TSMC 4N | TSMC 4N (4nm) |
| Release Date | 2022 3 GTC | 2024 |
| FP8 Compute | 3,958 TFLOPS | 296 TFLOPS |
| FP16 Compute | — | 148 TFLOPS |
| FP32 Compute | 67 TFLOPS | 24 TFLOPS |
| INT8 Compute | — | 296 TOPS |
| Memory Type | — | — |
| Memory Capacity | 80 GB HBM3 | — |
| Memory Bandwidth | 3.35 TB/s | 4.0 TB/s |
| TDP Power | 700 W | 400W |
Key Differences
- FP8 compute: NVIDIA H100 SXM leads with ~3,958 TFLOPS versus NVIDIA H20 (2024)'s 296 TFLOPS, a clear edge in large-scale Transformer training/inference.
- Power: NVIDIA H20 (2024) has a TDP of 400W, lower than NVIDIA H100 SXM's 700 W, friendlier to datacenter PUE and cooling.
Selection Advice
- When chasing extreme single-card compute and a mature toolchain, prioritize NVIDIA H100 SXM; if budget, power wall, or local support are hard constraints, NVIDIA H20 (2024) often fits better. Use this site's AI Compute Card Comparison Tool to validate multiple chips side-by-side before deciding.
FAQ
What are the main differences between NVIDIA H100 SXM and NVIDIA H20 (2024)?
The core difference is architecture and compute density: NVIDIA H100 SXM uses Hopper GH100, FP8 ~3,958 TFLOPS, memory 80 GB HBM3; NVIDIA H20 (2024) uses Hopper, FP8 ~296 TFLOPS, memory —. See the comparison table above.
What is the TDP (power) of NVIDIA H100 SXM?
NVIDIA H100 SXM has a TDP of 700 W; actual whole-system power also includes board, fans, and PUE.
Which is better for large-model training / inference?
Training values memory capacity, bandwidth, and multi-card interconnect; inference values single-card throughput and power efficiency. Combine the "Key Differences" and "Selection Advice" above with your batch size, model size, and SLA.
How much do NVIDIA H100 SXM and NVIDIA H20 (2024) differ in memory capacity?
NVIDIA H100 SXM is 80 GB HBM3, NVIDIA H20 (2024) is —; the gap directly affects loadable model size and context length.
Related Pages
- NVIDIA H100 SXM
- NVIDIA H20 (2024)
- AI 算力卡对比工具 — Compare 2–4 chips side-by-side online